ir>.S KNOTT AND TAX AK AD ATE 



seem l<> linvo kept fnirlv r-onstnnt. The several inensiiremenfs of tlie 

 ILn-izontal B'orce do ikH differ nmon_o-st themselves by so much ns 

 1 ill 1000; the Dips ao-ree to within 7'; and tlie mean monthly 

 deelinfitions to ^vithm o'. 



To reo'ard all the observations of the present snrvev as made at 

 the same epoch is. therefore, in the circumstances, the safest course. 

 Indeed it would be impossible with the data at our command to 

 apply monthly corrections ha^'ing any claim to probability. 



During the years 1882- o a magnetic survey of Ja])an was 

 carried out by IMessrs. Sekino and Kodari. members of the staff of the 

 Geological Survey Department. Their chart was jiresented at the 

 International Geological Congress at Berlin in 1885, and is reproduced, 

 along with the observations, in a pamphlet of Dr. Naumann's, entitled 

 "Die Erscheinungen der Erdmagnetismus in ihrer Abhängigkeit vom 

 r»au der Erdrinde" (Stuttgart, 1887). In this ]wmphlet, which 

 ('(mtaiiis a full account of previous magnetic work in Japan, Dr. 

 Naumann calls special attention to the apparent trend of the lines of 

 e(|ual magnetic declination, and suggests a possible relation ])etween 

 I heir f)rm and the «jeoloofical structure of the country. 



Some three or four year ago, however, an inspection of the results 

 of this first survey and a consideration of the routes chosen convinced 

 me that it would ])e unsafe to deduce from them any definite C(jn- 

 elusions as to the general magnetic characteristics of Japan. For 

 example, with the exception of a few in the neighliourhood of Niigata 

 and Sado island and a few between longitudes 136° and 138° E., 

 there were no stations along the westei-n coast of Japan. Nortli 

 of latitude 38° N. there was but a, single line of stations, confined 

 to tlie Ôshiû-kai-dô (the Öshiü High- way). Then the observations 

 were made in two sets, the one in 1882 between vVugiist 18th and 

 December 7th, and tlie other in 1883 between Se])te(nher 10th an<l 



